Our pervasive adoption of CMC in higher education has far outpaced our understanding of the nature of CMC and, accordingly, how this medium should best be used to promote higher-order learning (Garrison, Anderson, & Archer, "American Journal of Distance Education," 15 (1), 7-23, 2001). Therefore, this study examined the relationship between peer interaction and learning outcomes by creating and testing through structural equation modeling a model of five variables--learner characteristics measured by self-construal, perceived teaching style, task design, course requirements, and prior CMC experience--that influence self-reported peer interaction and self-reported learning outcomes. The results showed that perceived teaching style had a small direct effect (0.23) and prior CMC experience had a moderate direct effect (0.31) on self-reported peer interaction; self-reported peer interaction had a strong direct effect (0.66) on self-reported learning outcomes.